Summary of the Data

## [1] "total # of years"
## [1] 6
## [1] "total # of sites"
## [1] 17
## [1] "total # of site - year combinations"
## [1] 34
## [1] "total # of quadrats"
## [1] 960

Table S1

Site KI2013 KI2014 KI2015a KI2015b KI2016b KI2017
VL2 20 0 0 0 0 26
VL1 0 0 28 0 30 0
VL5 30 0 0 0 0 19
L5 25 0 0 0 0 29
M10 28 0 0 0 0 29
L1 0 32 0 0 0 29
H2 30 0 0 0 0 30
VH3 29 0 0 0 30 0
VH1 0 0 30 0 0 29
VH2 0 0 0 30 0 30
L4 25 0 0 0 0 30
M3 0 0 0 30 0 29
M2 26 0 0 0 30 0
M1 0 0 0 30 0 29
M4 0 21 0 0 0 30
M6 28 0 0 0 30 0
VL3 0 0 0 29 30 0

Coverage Standardizing

Figure S2

Hill Diversity

85% Coverage

Figure S4

90% Coverage

Figure 2

Figure S5

95% Coverage

Figure S6

Modelling

Table 1 and S3

  • The following models were used to create Table 1 and S3
## [1] "Hill-Richness"
##  Family: gaussian  ( identity )
## Formula:          qD ~ poly(HD_Cont, 2) * MHW + NPP + (1 | Site)
## Data: estimates_85 %>% filter(order == 0)
## 
##      AIC      BIC   logLik deviance df.resid 
##    142.1    155.9    -62.1    124.1       25 
## 
## Random effects:
## 
## Conditional model:
##  Groups   Name        Variance Std.Dev.
##  Site     (Intercept) 1.693    1.301   
##  Residual             1.126    1.061   
## Number of obs: 34, groups:  Site, 17
## 
## Dispersion estimate for gaussian family (sigma^2): 1.13 
## 
## Conditional model:
##                             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                22.009931   8.764425   2.511  0.01203 *  
## poly(HD_Cont, 2)1           3.098374   2.913080   1.064  0.28751    
## poly(HD_Cont, 2)2          -7.115000   2.395033  -2.971  0.00297 ** 
## MHWAfter                   -3.333235   0.363942  -9.159  < 2e-16 ***
## NPP                        -0.010909   0.008351  -1.306  0.19143    
## poly(HD_Cont, 2)1:MHWAfter -3.996851   2.122129  -1.883  0.05964 .  
## poly(HD_Cont, 2)2:MHWAfter -5.545241   2.122129  -2.613  0.00897 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## [1] "Hill-Shannon"
##  Family: gaussian  ( identity )
## Formula:          qD ~ poly(HD_Cont, 2) * MHW + NPP + (1 | Site)
## Data: estimates_85 %>% filter(order == 1)
## 
##      AIC      BIC   logLik deviance df.resid 
##    137.9    151.6    -59.9    119.9       25 
## 
## Random effects:
## 
## Conditional model:
##  Groups   Name        Variance Std.Dev.
##  Site     (Intercept) 0.6442   0.8026  
##  Residual             1.4467   1.2028  
## Number of obs: 34, groups:  Site, 17
## 
## Dispersion estimate for gaussian family (sigma^2): 1.45 
## 
## Conditional model:
##                             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                19.020068   6.825650   2.787 0.005327 ** 
## poly(HD_Cont, 2)1           1.590186   2.430764   0.654 0.512988    
## poly(HD_Cont, 2)2          -6.996675   2.059513  -3.397 0.000681 ***
## MHWAfter                   -3.953707   0.412555  -9.583  < 2e-16 ***
## NPP                        -0.009310   0.006502  -1.432 0.152173    
## poly(HD_Cont, 2)1:MHWAfter -1.327062   2.405589  -0.552 0.581183    
## poly(HD_Cont, 2)2:MHWAfter -1.647390   2.405589  -0.685 0.493459    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## [1] "Hill-Simpson"
##  Family: gaussian  ( identity )
## Formula:          qD ~ poly(HD_Cont, 2) * MHW + NPP + (1 | Site)
## Data: estimates_85 %>% filter(order == 2)
## 
##      AIC      BIC   logLik deviance df.resid 
##    132.9    146.6    -57.4    114.9       25 
## 
## Random effects:
## 
## Conditional model:
##  Groups   Name        Variance Std.Dev.
##  Site     (Intercept) 0.2001   0.4473  
##  Residual             1.5285   1.2363  
## Number of obs: 34, groups:  Site, 17
## 
## Dispersion estimate for gaussian family (sigma^2): 1.53 
## 
## Conditional model:
##                             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                15.863242   5.733000   2.767 0.005657 ** 
## poly(HD_Cont, 2)1           0.461803   2.162136   0.214 0.830869    
## poly(HD_Cont, 2)2          -6.727069   1.870653  -3.596 0.000323 ***
## MHWAfter                   -4.128235   0.424063  -9.735  < 2e-16 ***
## NPP                        -0.007191   0.005460  -1.317 0.187788    
## poly(HD_Cont, 2)1:MHWAfter  0.015729   2.472689   0.006 0.994924    
## poly(HD_Cont, 2)2:MHWAfter  0.488883   2.472689   0.198 0.843270    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## [1] "Hill-Richness"
##  Family: gaussian  ( identity )
## Formula:          qD ~ poly(HD_Cont, 2) * MHW + NPP + (1 | Site)
## Data: estimates_90 %>% filter(order == 0)
## 
##      AIC      BIC   logLik deviance df.resid 
##    159.5    173.2    -70.7    141.5       25 
## 
## Random effects:
## 
## Conditional model:
##  Groups   Name        Variance Std.Dev.
##  Site     (Intercept) 2.235    1.495   
##  Residual             2.134    1.461   
## Number of obs: 34, groups:  Site, 17
## 
## Dispersion estimate for gaussian family (sigma^2): 2.13 
## 
## Conditional model:
##                            Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                27.48037   10.60354   2.592  0.00955 ** 
## poly(HD_Cont, 2)1           2.80018    3.59242   0.779  0.43570    
## poly(HD_Cont, 2)2          -7.76270    2.98011  -2.605  0.00919 ** 
## MHWAfter                   -3.36594    0.50103  -6.718 1.84e-11 ***
## NPP                        -0.01449    0.01010  -1.434  0.15143    
## poly(HD_Cont, 2)1:MHWAfter -3.80986    2.92146  -1.304  0.19220    
## poly(HD_Cont, 2)2:MHWAfter -8.97593    2.92146  -3.072  0.00212 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## [1] "Hill-Shannon"
##  Family: gaussian  ( identity )
## Formula:          qD ~ poly(HD_Cont, 2) * MHW + NPP + (1 | Site)
## Data: estimates_90 %>% filter(order == 1)
## 
##      AIC      BIC   logLik deviance df.resid 
##    141.8    155.6    -61.9    123.8       25 
## 
## Random effects:
## 
## Conditional model:
##  Groups   Name        Variance Std.Dev.
##  Site     (Intercept) 0.7324   0.8558  
##  Residual             1.6190   1.2724  
## Number of obs: 34, groups:  Site, 17
## 
## Dispersion estimate for gaussian family (sigma^2): 1.62 
## 
## Conditional model:
##                             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                21.193684   7.247612   2.924 0.003453 ** 
## poly(HD_Cont, 2)1           1.549540   2.578689   0.601 0.547905    
## poly(HD_Cont, 2)2          -7.755289   2.184061  -3.551 0.000384 ***
## MHWAfter                   -4.303176   0.436427  -9.860  < 2e-16 ***
## NPP                        -0.010518   0.006904  -1.524 0.127621    
## poly(HD_Cont, 2)1:MHWAfter -1.310906   2.544784  -0.515 0.606459    
## poly(HD_Cont, 2)2:MHWAfter -2.030854   2.544784  -0.798 0.424844    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## [1] "Hill-Simpson"
##  Family: gaussian  ( identity )
## Formula:          qD ~ poly(HD_Cont, 2) * MHW + NPP + (1 | Site)
## Data: estimates_90 %>% filter(order == 2)
## 
##      AIC      BIC   logLik deviance df.resid 
##    137.3    151.0    -59.6    119.3       25 
## 
## Random effects:
## 
## Conditional model:
##  Groups   Name        Variance Std.Dev.
##  Site     (Intercept) 0.1424   0.3773  
##  Residual             1.8179   1.3483  
## Number of obs: 34, groups:  Site, 17
## 
## Dispersion estimate for gaussian family (sigma^2): 1.82 
## 
## Conditional model:
##                             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                16.723613   5.986392   2.794 0.005212 ** 
## poly(HD_Cont, 2)1           0.484751   2.290866   0.212 0.832418    
## poly(HD_Cont, 2)2          -7.387219   1.991615  -3.709 0.000208 ***
## MHWAfter                   -4.455059   0.462466  -9.633  < 2e-16 ***
## NPP                        -0.007462   0.005701  -1.309 0.190565    
## poly(HD_Cont, 2)1:MHWAfter -0.073007   2.696620  -0.027 0.978401    
## poly(HD_Cont, 2)2:MHWAfter  0.790979   2.696620   0.293 0.769276    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## [1] "Hill-Richness"
##  Family: gaussian  ( identity )
## Formula:          qD ~ poly(HD_Cont, 2) * MHW + NPP + (1 | Site)
## Data: estimates_95 %>% filter(order == 0)
## 
##      AIC      BIC   logLik deviance df.resid 
##    190.4    204.2    -86.2    172.4       25 
## 
## Random effects:
## 
## Conditional model:
##  Groups   Name        Variance Std.Dev.
##  Site     (Intercept) 0.997    0.9985  
##  Residual             8.388    2.8962  
## Number of obs: 34, groups:  Site, 17
## 
## Dispersion estimate for gaussian family (sigma^2): 8.39 
## 
## Conditional model:
##                             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                 33.20798   13.30160   2.497 0.012541 *  
## poly(HD_Cont, 2)1            0.01762    5.03238   0.004 0.997207    
## poly(HD_Cont, 2)2           -8.18759    4.35856  -1.879 0.060312 .  
## MHWAfter                    -3.73394    0.99339  -3.759 0.000171 ***
## NPP                         -0.01732    0.01267  -1.367 0.171497    
## poly(HD_Cont, 2)1:MHWAfter   1.04902    5.79243   0.181 0.856287    
## poly(HD_Cont, 2)2:MHWAfter -15.73564    5.79243  -2.717 0.006596 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## [1] "Hill-Shannon"
##  Family: gaussian  ( identity )
## Formula:          qD ~ poly(HD_Cont, 2) * MHW + NPP + (1 | Site)
## Data: estimates_95 %>% filter(order == 1)
## 
##      AIC      BIC   logLik deviance df.resid 
##    146.2    159.9    -64.1    128.2       25 
## 
## Random effects:
## 
## Conditional model:
##  Groups   Name        Variance Std.Dev.
##  Site     (Intercept) 0.7507   0.8664  
##  Residual             1.8970   1.3773  
## Number of obs: 34, groups:  Site, 17
## 
## Dispersion estimate for gaussian family (sigma^2):  1.9 
## 
## Conditional model:
##                             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                22.378736   7.608535   2.941 0.003269 ** 
## poly(HD_Cont, 2)1           1.092166   2.727785   0.400 0.688873    
## poly(HD_Cont, 2)2          -8.269011   2.317231  -3.568 0.000359 ***
## MHWAfter                   -4.730051   0.472410 -10.013  < 2e-16 ***
## NPP                        -0.010678   0.007247  -1.473 0.140651    
## poly(HD_Cont, 2)1:MHWAfter -0.575387   2.754601  -0.209 0.834540    
## poly(HD_Cont, 2)2:MHWAfter -3.026929   2.754601  -1.099 0.271828    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## [1] "Hill-Simpson"
##  Family: gaussian  ( identity )
## Formula:          qD ~ poly(HD_Cont, 2) * MHW + NPP + (1 | Site)
## Data: estimates_95 %>% filter(order == 2)
## 
##      AIC      BIC   logLik deviance df.resid 
##    142.0    155.8    -62.0    124.0       25 
## 
## Random effects:
## 
## Conditional model:
##  Groups   Name        Variance Std.Dev.
##  Site     (Intercept) 0.08229  0.2869  
##  Residual             2.16806  1.4724  
## Number of obs: 34, groups:  Site, 17
## 
## Dispersion estimate for gaussian family (sigma^2): 2.17 
## 
## Conditional model:
##                             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                16.839739   6.305676   2.671 0.007572 ** 
## poly(HD_Cont, 2)1           0.261876   2.444057   0.107 0.914671    
## poly(HD_Cont, 2)2          -7.884811   2.133469  -3.696 0.000219 ***
## MHWAfter                   -4.790117   0.505041  -9.485  < 2e-16 ***
## NPP                        -0.007039   0.006004  -1.172 0.241097    
## poly(HD_Cont, 2)1:MHWAfter  0.181635   2.944868   0.062 0.950819    
## poly(HD_Cont, 2)2:MHWAfter  0.878027   2.944868   0.298 0.765585    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Average Losses

Table S4

## # A tibble: 1 × 2
##   HillRichness SEM_HillRichness
##          <dbl>            <dbl>
## 1         3.37            0.664
## # A tibble: 1 × 2
##   HillShannon SEM_HillShannon
##         <dbl>           <dbl>
## 1        4.30           0.462
## # A tibble: 1 × 2
##   HillSimpson SEM_HillSimpson
##         <dbl>           <dbl>
## 1        4.46           0.478
HD_Cat HillRichness SEM_HillRichness HillShannon SEM_HillShannon HillSimpson SEM_HillSimpson
Low 1.991667 1.9410242 4.628667 1.3880636 5.334333 1.3185235
Medium 1.653500 1.0039179 3.679833 0.9446390 4.297167 0.9973579
Very High 5.862250 0.8116662 4.997750 0.4508515 4.390250 0.7816353
Very Low 4.469000 0.7238535 4.299500 1.0254281 4.097250 0.9799411

Stressor Responses

Note: Notation in code is the same as the notation used in the manuscript equations. ## Figure 4 (a-c) ### (a) Richness

## 
##  supF test
## 
## data:  fs.AR_Richness
## sup.F = 12.429, p-value = 0.03151
## 
##   Optimal 2-segment partition: 
## 
## Call:
## breakpoints.Fstats(obj = fs.AR_Richness)
## 
## Breakpoints at observation number:
## 9 
## 
## Corresponding to breakdates:
## 0.4705882
## 
## Call:
## lm(formula = ARi ~ HD_Cont, data = subset(AR_Richness, HD_Cont <= 
##     34.82568))
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.31229 -0.10883 -0.04707  0.06764  0.39817 
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)
## (Intercept)  0.004617   0.116360   0.040    0.970
## HD_Cont     -0.008649   0.005835  -1.482    0.189
## 
## Residual standard error: 0.2363 on 6 degrees of freedom
## Multiple R-squared:  0.268,  Adjusted R-squared:  0.146 
## F-statistic: 2.197 on 1 and 6 DF,  p-value: 0.1888
## 
## Call:
## lm(formula = ARi ~ HD_Cont, data = subset(AR_Richness, HD_Cont >= 
##     34.82568))
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.41895 -0.08000  0.01754  0.11164  0.33289 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)  
## (Intercept) -0.61544    0.22569  -2.727   0.0295 *
## HD_Cont      0.01097    0.00419   2.618   0.0345 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.2296 on 7 degrees of freedom
## Multiple R-squared:  0.4947, Adjusted R-squared:  0.4225 
## F-statistic: 6.854 on 1 and 7 DF,  p-value: 0.03451

(b) Shannon

## 
##  supF test
## 
## data:  fs.AR_Shannon
## sup.F = 3.939, p-value = 0.7032
## 
##   Optimal 2-segment partition: 
## 
## Call:
## breakpoints.Fstats(obj = fs.AR_Shannon)
## 
## Breakpoints at observation number:
## 6 
## 
## Corresponding to breakdates:
## 0.2941176
## 
## Call:
## lm(formula = ARi ~ HD_Cont, data = AR_Shannon)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.52140 -0.10413  0.00812  0.10876  0.40314 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.012963   0.098560   0.132    0.897
## HD_Cont     0.001128   0.002374   0.475    0.642
## 
## Residual standard error: 0.2419 on 15 degrees of freedom
## Multiple R-squared:  0.01482,    Adjusted R-squared:  -0.05086 
## F-statistic: 0.2257 on 1 and 15 DF,  p-value: 0.6416

(c) Simpson

## 
##  supF test
## 
## data:  fs.AR_Simpson
## sup.F = 2.872, p-value = 0.88
## 
##   Optimal 2-segment partition: 
## 
## Call:
## breakpoints.Fstats(obj = fs.AR_Simpson)
## 
## Breakpoints at observation number:
## 6 
## 
## Corresponding to breakdates:
## 0.2941176
## 
## Call:
## lm(formula = ARi ~ HD_Cont, data = AR_Simpson)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.51523 -0.24739  0.06136  0.12391  0.48950 
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.030e-01  1.175e-01   0.877    0.395
## HD_Cont     7.178e-05  2.830e-03   0.025    0.980
## 
## Residual standard error: 0.2883 on 15 degrees of freedom
## Multiple R-squared:  4.289e-05,  Adjusted R-squared:  -0.06662 
## F-statistic: 0.0006433 on 1 and 15 DF,  p-value: 0.9801